Figuring out how to automate voltage measurements

[rx8pilot]

I am trying to automate some of the testing I have been doing manually on PCB's as they are being built. This is for a low-volume operation but the tests are taking a lot of time since each one has a lot of points that need to be verified for voltage. There are other tests that are being done, but for the sake of keeping this discussion reasonable - I am only looking at practical ways to measure and log about 20-30 points for DC voltages.

Setup:
I have been using LabView - seems really great so far. Tried Python and although I could eventually get it done, it is slow to dev and maintain. In the end, the solution will be MUCH more than capturing a few dozen voltages so LabView seemed like a good choice.

I have been manually probing but in the process of putting together a bed of nails for all the PCB's involved (3 at the moment). They are double sided PCBs and the bed of nails is double sided.

The goal is to use relays or other switching so I can use a single DMM to do the measurements. I don't really know where to start. So far, I looked at Keithley DMM with data acquisition slots - the 2700 series. I found a few GPIB relay boxes. I looked at building a PXI chassis with National Instruments DMM's. I saw some PXI relay cards.

Has anyone done tests like this before? Straight DC, multi-channel automated tests? Are relays the way to go to feed a single meter?

[pickle9000]

- What kind of precision do you need?
- How extensive are these tests (will there be final product testing)?
- How much do you want to spend?
- Will you be tearing your fixtures down after testing?

[rx8pilot]

I would say 3 decimal places is acceptable. These tests are for final product releases to detect anomalies from manufacturing and build calibration tables for the microcontrollers.

My guess is that I need to spend at least few thousand from what I  have seen. The bed of nails not included since that is moving forward anyway.

The fixtures will be cartridges that go into a general purpose double-sided bed of nails. I was thinking about using a DB50 or similar for DC and SMA's for high frequency. Those will hopefully allow easy changeover for the different PCB's.

Sent from my SM-T800 using Tapatalk

[nfmax]

Have you looked at the Keysight/Agilent/HP 34970A or 34972A? With a single 34901A module they give you 20 fully isolated channels: they hold up to 3 modules. The DVM part is basically a 34401A. They turn up regularly on eBay.

[rx8pilot]

I have not seen those before, but seem like a clean all-in-one solution. I did, however, see the Keithley 2700's which seem to be a similar box but has more functionality as a normal stand-alone DMM. The 2700 seems like it is lower cost (at least in the used market) but the relay cards a much more expensive than Agilent.

[rx8pilot]

Well, I managed to get a Keithley 2700 auction on eBay for a great price. Let's see how it works when it arrives - the pics look ok but the description is non-existent. I plan to do the basic run down and then take it in for calibration before I do anything else. There are dozens of labs close by.

I like that the 2700 can work as a normal every day 6.5 DMM while having the automation slots in the back. Once it gets checked out, I will be on the hunt for the solid state relay module for 20 channels of measurments.

Hope this is a good choice. "production" instruments and testing techniques are totally new to me.